Buying Movement Itself
In the last week of January 2018, one-month implied volatility on the S&P 500 sat near 11%. Over the following two weeks the index moved more than 2% on four separate days, including a 4.1% drop on 5 February, and realised volatility exploded past 30%. Anyone who had bought options at that 11% implied and hedged the direction away did not need to predict where the market was going. They were paid, handsomely, for the simple fact that it moved.
That trade has a name: gamma scalping. You buy options, remove the directional exposure by delta hedging, and then mechanically trade the underlying against the option position as the market wobbles. Done long enough, the P&L stops being about direction at all and becomes a clean bet on one number: whether the volatility the market realises exceeds the volatility you paid for.
Most of the time it does not. Carr and Wu (2009) documented that index implied volatility carries a persistent premium over subsequent realised volatility, which is exactly why selling volatility is a business. Gamma scalping is the other side of that business, and understanding when it pays is understanding when everyone else's insurance premium was too cheap.
The Setup
The standard vehicle is a long at-the-money straddle: buy a call and a put on the same strike and expiry. The position starts roughly delta neutral, has positive gamma (its delta rises as the underlying rises and falls as it falls) and negative theta (it loses value each day as the options decay). Gamma and theta are the two Greeks that run this entire strategy; everything else is detail.
Positive gamma has a lovely mechanical consequence. When the stock rallies, your delta becomes positive, so you sell stock to get back to neutral. When the stock drops, your delta goes negative, so you buy. The hedging rule forces you to sell high and buy low, automatically, every time you rebalance. Each round trip locks in a small profit that exists only because the underlying moved.
Those hedging profits are the "scalps". Against them runs the theta bill: the daily cost of holding decaying options. The whole strategy is a race between the two.
The Mechanics, One Rebalance at a Time
Make it concrete. A stock trades at 100. You buy straddles with a combined position gamma of 600 shares per point (meaning a $1 move shifts your delta by about 600 shares) and total theta of $900 per day. You start delta flat.
Day one. The stock rallies from 100 to 102. With gamma of 600 per point, your delta has climbed to roughly +1,200 shares. You sell 1,200 shares at 102 to flatten. The profit from gamma over that move is approximately:
gamma P&L = 0.5 * gamma * (move)^2 = 0.5 * 600 * 2^2 = $1,200
Theta costs $900. Net for the day: +$300.
Day two. The stock drifts from 102 to 101.50. Gamma P&L = 0.5 * 600 * 0.5^2 = $75. Theta again takes $900. Net: -$825. Quiet days bleed.
Day three. The stock gaps down from 101.50 to 98.50, a $3 move. Your delta is now about -1,800 shares; you buy 1,800 shares at 98.50. Gamma P&L = 0.5 * 600 * 9 = $2,700 against $900 of theta. Net: +$1,800.
Three days in, you are up $1,275 without ever holding a directional view for more than a few hours. Notice the shape of the arithmetic: gamma profits grow with the square of the move, theta is a flat daily toll. Big-move days pay for many quiet ones.
The breakeven daily move for this position is the move where gamma income equals theta: 0.5 * 600 * m^2 = 900 gives m = $1.73, about 1.73% of a $100 stock. Annualised (multiply by the square root of 252), that is roughly 27% volatility. Which is, not coincidentally, close to what the options were implying when you bought them. The market quotes you a breakeven; gamma scalping is the wager that reality beats it.
The P&L Identity
The example generalises into one of the most useful identities in derivatives. For a delta-hedged option position, the profit over a short interval is approximately:
P&L = 0.5 * gamma * S^2 * (realised variance - implied variance) * dt
where S is the underlying price and the variances are the realised move actually delivered versus the move implied by the volatility you paid. The term 0.5 * gamma * S^2 is dollar gamma; the bracket is the entire game. Hedge a long option to expiry and your total P&L is, cost frictions aside, a dollar-gamma-weighted average of (realised minus implied) variance along the path.
Three consequences follow. First, gamma scalping is not an income strategy or a pattern to spot; it is the physical delivery mechanism of a volatility bet. Second, the weighting matters: gamma is largest at the money and near expiry, so realised volatility that arrives while the spot is near your strike counts far more than the same volatility delivered after the underlying has drifted away. Third, direction truly cancels: an up move and a down move of equal size are worth exactly the same.
Buying volatility through implied volatility below 11% in late January 2018 worked not because anyone called the crash, but because realised variance came in at multiples of what was paid.
Simulating the Race in Python
A ten-line simulation makes the gamma-versus-theta race visible. This holds dollar gamma and theta fixed for clarity; in reality both drift as spot and time move.
import numpy as np rng = np.random.default_rng(7) S = 100.0 # spot gamma = 600.0 # shares of delta per 1 point move theta = 900.0 # dollars lost per day implied_be = np.sqrt(2 * theta / gamma) # daily breakeven move for realised_vol in (0.15, 0.27, 0.40): moves = rng.normal(0, realised_vol / np.sqrt(252), 21) * S gamma_pnl = 0.5 * gamma * moves**2 net = gamma_pnl.sum() - theta * 21 print(f"realised {realised_vol:.0%}: net P&L over " f"21 days = {net:>8,.0f} (breakeven move {implied_be:.2f})")
Run it and the pattern is stark: at 15% realised the position bleeds steadily, near 27% it roughly washes, at 40% it prints. The randomness within each regime is also instructive: even at high realised volatility, an unlucky path with the moves clustered on days you were poorly positioned produces mediocre results. Path dependence is a feature of the trade, not a bug in the simulation.
Who Actually Does This
Options market makers are the largest gamma scalpers on earth, mostly by necessity. A market-making desk that ends the day long options from customer selling flow is long gamma whether it likes it or not, and hedging that book is gamma scalping in industrial form. Whether the desk keeps the gamma or flattens it is a live pricing decision about realised versus implied volatility.
Volatility arbitrage funds run it deliberately: buy options where implied looks cheap relative to forecast realised, sell where rich, delta-hedge everything, and let the identity above convert the forecast into P&L. The skill sits in the volatility forecast and in execution, not in the hedging mechanics, which are commoditised.
Dispersion and relative-value desks use gamma scalping as the engine inside larger structures, long single-name gamma against short index gamma, where each leg is delta-hedged and the bet is about which realised volatility outruns its implied.
Practical Questions
How often should you rebalance? There is no free answer. Hedging on a fixed clock (say daily) is simple but lets delta drift between hedges; hedging on delta bands (rebalance whenever delta exceeds some threshold) adapts to conditions and is what most practitioners use. Rebalancing more often reduces the variance of your P&L but not its expectation, while every trade pays the bid-ask spread. The classic result is that the optimal frequency scales with transaction costs: hedge tighter when trading is cheap, looser when it is dear.
What do costs do to the trade? They subtract a spread-crossing toll from every scalp, which hits exactly the small frequent rebalances that quiet markets produce. A strategy that looks marginally profitable at mid prices is often a steady loser after costs. Liquid underlyings with tight spreads and cheap options commissions are close to a precondition.
Cash or futures? Hedging with futures where available is usually cheaper and cleaner than trading the cash underlying, particularly for index positions.
When Gamma Scalping Loses
The failure modes are well charted, and the first one is the default state of the world.
Realised comes in below implied. This is the base case, not the exception. The variance risk premium exists because option sellers demand compensation for crash risk, so buying gamma at the market's asking price and hedging it is, on average and over long samples, a losing trade on indices. Systematic long-gamma programmes survive only with a genuine forecasting edge or as hedges whose negative carry is accepted for their crisis payoff.
Theta bleed in dead markets. The losses do not arrive as one dramatic event but as a fortnight of $825 days while the underlying goes nowhere. Long gamma positions have to be sized so that a quiet month is an annoyance rather than a forced exit at the lows of implied.
Gaps you cannot scalp. The identity assumes you can rebalance along the path. An overnight gap from 100 to 90 pays you the full gamma profit on a 10-point move, which sounds ideal until the position is short gamma, in which case there was no chance to hedge anything on the way down. For the long-gamma trader gaps are the best days; the mirror-image lesson is that short-gamma scalping profits are an advance on losses that arrive all at once.
Weekends and holidays. Theta accrues over calendar days; scalping opportunities only occur on trading days. Holding long gamma into a three-day weekend means paying three days of decay for zero rebalancing opportunities, which is why implied volatility routinely gets marked down into long weekends and why Friday-afternoon gamma is famously cheap for a reason.
Gamma scalping is best understood not as a strategy to deploy but as a lens: it is what a volatility view looks like once every other exposure has been hedged away. If you cannot say why realised volatility should beat the implied you are paying, the machinery will execute your lack of a view flawlessly.
Skip the £25k programme - try the alternative
Master's programmes are slow and expensive. Quantt is a self-paced alternative. Start free with a real lesson and interview practice, then unlock 50+ courses and your personalised plan.
Free to start · No credit card required